Вы находитесь на странице: 1из 21

A Sign of Things to Come: Behavioral

Change through Dynamic Iconography


n a daily basis we consume countless public and

private facilities, from swimming pools to beaches,
neighborhood playgrounds to amusement parks, and parking lots to roads. While it is important to study what facilities we consume, it is also important to study how we
consume them. Irresponsible consumption, such as ignoring warning signs and product labels, leads to costs that
are incurred not only by the consumer and the company
but also by society as a whole. For example, over 60,000
people each year require emergency room treatment for
injuries sustained in swimming pool accidents in the

United States (Kennerly 2014). Many of these accidents

are due to negligent behavior and inadequate warning signs
(Dworkin 2014). Responsible consumption, therefore, is
not only desirable but also extremely important. One component of responsible consumption is following posted
rules and signs. For example, if one is at the zoo, one
should not get too close to the animals. When swimming
at a beach, one should avoid areas that contain jellyfish.
When working with products containing acid, one should
be mindful of the possible consequences (see fig. 1).
Let us consider just one domain in which warning signs
play an important role: driving a vehicle. Every day, throughout the world, there are 3,287 deaths due to car accidents. In
the United States alone, 37,000 people die in road crashes
each year, with an additional 2.35 million people being either
injured or disabled (ASIRT 2014). These terrible consequences are not limited to just drivers or passengers in cars.
Nearly one in every five children between the ages of 5 and
9 killed in traffic accidents was a pedestrian. In addition, of
the 6.3 million car accidents recorded annually in the United
States, 1.26 million occur in parking lots (AAC 2010).
In an effort to reduce the number of accidents, businesses
and governments worldwide place traffic signs in areas of
potential danger. Children are taught from a young age to
abide by these traffic signs. Motorists learn about the signs,
are tested for their comprehension of the signs, and are fined
for any violations of the signal-suggested behavior. Interestingly, there is considerable variance both within and

Luca Cian (luca_cian@yahoo.it) is assistant professor of marketing at

the Darden School of Business, University of Virginia, Charlottesville, VA
22903. Aradhna Krishna (aradhna@umich.edu) is the Dwight F. Benton
Professor of Marketing at the Ross School of Business, University of
Michigan, Ann Arbor, MI 48109. Ryan S. Elder (rselder@byu.edu) is
assistant professor of marketing at the Marriott School of Management,
Brigham Young University, Provo, UT 84602. The first two authors contributed equally to the research, and their names are in alphabetical order.
This research was conducted when the first author was a postdoctoral
scholar at Aradhna Krishnas Sensory Marketing Lab, University of Michigan. The authors acknowledge the valuable input of the editor, the associate
editor, and the reviewers. Correspondence: Luca Cian and Aradhna
Laura Peracchio and Vicki Morwitz served as editors and Rashmi Adaval
served as associate editor for this article.
Electronically published February 17, 2015

2015 by JOURNAL OF CONSUMER RESEARCH, Inc. Vol. 41 April 2015
All rights reserved. 0093-5301/2015/4106-0006$10.00. DOI: 10.1086/680673

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

We propose that features of static visuals can lead to perceived movement (via
dynamic imagery) and prepare the observer for action. We operationalize our
research within the context of warning sign icons and show how subtle differences
in iconography can affect human behavioral response. Across five studies incorporating multiple methodologies and technologies (click-data heat maps, driving
simulations, surveys, reaction time, and eye tracking), we show that warning sign
icons that evoke more (vs. less) perceived movement lead to a quicker propensity
to act because they suggest greater risk to oneself or others and increase attentional vigilance. Icons used in our studies include children crossing signs near
schools, wet floor signs in store settings, and shopping cart crossings near malls.
Our findings highlight the importance of incorporating dynamic elements into icon
design to promote imagery and thereby elicit desired and responsible consumer



spective, our research is important because it adds to the

literature on image attributes that facilitate imagery (e.g.,
Adaval and Wyer 1998; MacInnis and Price 1987; Petrova
and Cialdini 2005; Wang and Peracchio 2007) by highlighting the dynamic nature of imagery and exploring its consequences on behavior. To explore both these substantive
and theoretical implications, we focus on driving and traffic
signs. Surprisingly, despite consumers spending an average
of over 2 hours per day in their cars (Arbitron, Edison Media
Research 2013), very little research in consumer psychology
has focused on the driving context or on driving behaviors
(for an exception, see Wood, McInnes, and Norton 2011).
This domain, therefore, provides a rich avenue within which
to conduct our research.
The rest of the article is organized as follows. Next we
discuss literature pertinent to our research and build our
conceptual model. This is followed by a description of five
studies and their findings. We conclude with implications,
limitations, and possible extensions of our research.


Literature relevant to our research is found in the areas
of visual imagery, dynamic imagery, and completion of
movement; perceived movement and attention; and per-


Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

across countries for traffic icons designed to elicit the very

same behavior. Figure 2 shows the traffic icon for a school
crossing where motorists are expected to slow down when
they see the sign. We note differences in traffic icons across
the United States, Poland, and Russia, while icons in other
countries bring additional variations. We focus on one dimension of the iconthe perceived movement (or dynamic
imagery) that the icon elicits. All three icons in figure 2 are
static visuals (i.e., they have no animation). Yet, each of
them has some element of dynamism, with the Russian
icon appearing most dynamic, evoking the highest perceived
movement of the children crossing. Can the way the traffic
signs are designed impact desired behavior?
The design of warning signs and its effects on behavior
have both substantive and theoretical implications. The substantive implications are particularly important given the
implications warning signs have for human life as well as
because of the significant costs that are likely to be incurred
by companies and the government when they are not followed. A recent call for research on consumer and societal
well-being (Mari 2008; Mick 2008; Mick et al. 2011) reflects
the need for more research on how responsible consumption
can be encouraged. In the context of warning signs, it is
prudent to consider how best to design static warning signs
to effect behavior change given that animation is typically
expensive and not always doable. From a theoretical per-


ceived movement and human behavior. We briefly discuss

each in turn.

Dynamic Visual Imagery and the Completion

of Movement

midst of motion is captured (e.g., a still photograph of an

individual jumping in mid-air or the trajectory of a ball
in flight; Freyd 1983). In these snap shots, it has been
shown that one can visualize movementfor example, the
individual and the ball coming back to the ground in the
actions cited earlier. Neuroscientific evidence also supports
such dynamic imagery. The same brain regions (the medial
temporal/medial superior temporal cortex; MT/MST) are
active when observing action and when imagining motion
(Goebel et al. 1998; OCraven and Kanwisher 1997). Importantly, these brain regions are also active when viewing
static images where motion is implied (Kourtzi and Kanwisher 2000), suggesting that even in the absence of actual
motion, the brain is perceiving movement. While the bulk
of the prior literature has focused on when dynamic imagery and perceived movement occur, recent research in
consumer behavior has shown that perceived movement
has consequences not only on memory but also on consumer evaluations. For example, using brand logos with
lower and higher dynamic imagery, Cian, Krishna, and
Elder (2014) show that more (vs. less) dynamic brand logos
enhance brand evaluation.
To summarize, findings from existing research suggest
that (a) people can spontaneously generate mental images
and (b) static visual stimuli can lead to perceived movement,
with dynamic imagery on the part of the observer leading
to mental completion of the movement in ones mind. In
this research, we operationalize dynamic imagery within the
context of warning sign icons and explore if dynamism
within these icons can stimulate people to imagine how
objects move. This perception of movement should also lead
to behavioral consequences, which were not explored in the
prior literature. Previous research has shown that the completion of movement in the mind corresponds to a more
vivid image of the event depicted (Callow, Roberts, and
Fawkes 2006; Smallwood et al. 2004). Thus, increases in
dynamism to a sign icon depicting a dangerous situation
should lead to a more vivid image of the action depicted.
A vivid representation of a dangerous situation (e.g., a rock
falling on the road) should then lead to heightened levels
of perceived risk as imagining oneself in a scenario increases
its perceived likelihood of occurring (Carroll 1978; MacInnis and Price 1987).

Perceived Movement, Risk Perception, and

Attentional Vigilance
As humans have evolved, they have learned to be vigilant
to threat. Motion plays a large role in determining what people
attend to and how vigilant they are in their attentional response.
In an article on the effects of animacy (the life and motion
perceived in an object) on attentional capture, Pratt et al. (2010)
highlighted how animacy affects prioritization in visual processing. Indeed, humans have developed systems to maximize
the chances of detecting potential predators and other dangers,
one of these systems being animacy detection (Barrett 2005).
Animacy signals something unpredictable that one might have

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

Visual imagery has been studied extensively within the

psychology and consumer behavior literatures, providing a
foundational framework for not only what facilitates image
generation but also for the resulting content of these images
(see Kosslyn, Thompson, and Ganis 2006; Petrova and Cialdini 2008). The elicitation of visual images occurs at both
deliberate and spontaneous levels. At a deliberate level, visual images may be formed as a consequence of explicit
instructions to imagine (e.g., Gregory, Cialdini, and Carpenter 1982; McGill and Anand 1989; Morewedge, Huh,
and Vosgerau 2010). In addition, and importantly for our
conceptual model, visual images can also be formed spontaneously. For instance, reading a concrete word or description can not only elicit an image (Paivio 1969) but also lead
to automatic neural activity in corresponding sensory
regions of the brain (Gonzalez et al. 2006). Pictures alone
can also lead to spontaneous generation of imagery, with
vividness (Petrova and Cialdini 2005), stimulus orientation
(Elder and Krishna 2012), and even stimulus type (e.g., hand
tool vs. not; Chao and Martin 2000) affecting the amount
of imagery generated. The content of these images, in addition to how the images are formed, plays a large role in
affecting evaluations and behavior.
We explore how perception of motion from a static picture
spontaneously facilitates dynamic imagery, with people
mentally completing the movement of the action depicted
in the picture. Images are not restricted to static pictures in
the minds eye but can also contain a visual representation
of movement of the stimulus or even the self. In early research exploring the notion of kinematics (motion) and dynamics (motion and also the forces that produce motion),
participants showed the ability to imagine the continuation
of movement once movement of a stimulus stops, as well
as the movement of shapes from a still line drawing (Anderson, Howe, and Tolmie 1996). These effects obtain as
our minds draw upon past experiences in memory to continue the motion that we have perceived initially (McCloskey 1983). Such dynamic imagery has been studied most
extensively within the rubric of Representational Momentum (RM; Freyd and Finke 1984). A typical study exploring
this phenomenon has participants view objects that move
on a computer screen and then disappear. Participants are
then asked to identify where the object disappeared on the
axis of motion. The common finding is that participants
remember the object to have disappeared farther along the
axis of motion than the actual vanishing point (e.g., Hubbard
and Bharucha 1988).
The vast majority of RM studies utilize moving objects
as described. However, some research investigates perceived motion using static images. Most commonly these
are frozen-action images, in which a moving figure in the



H1a: A warning sign icon (e.g., a yield sign) with more

(vs. less) perceived movement will attract earlier
H1b: A warning sign icon with more (vs. less) perceived movement will evoke greater attentional

Perceived Movement and Human Behavior

Much recent research in situated cognition suggests that
thinking is for doing and facilitating action in a manner that
considers the current situation (Smith and Semin 2004,
2007). Because thinking is for doing and is influenced by
the settings of the immediate context, the particular actions
or thoughts that are activated are a function of what is psychologically relevant and salient in the context, together with
situational action potentials (Oyserman 2009). In particular,
the mental representation of a concept prepares the individual for action in the context the concept is placed in. This
includes the setting, actions, and introspections that go along
with the concept, creating the experience of being there

(Barsalou 2008; Niedenthal et al. 2005). Consequently, perceived movement from static images can not only activate
the neural structures involved in motor planning and execution for the corresponding movement but also lead to a
more general preparation to act in the given context. Because perception of movement prepares one for actual movement and a warning sign icon with more (vs. less) perceived
movement should result in earlier attention and increased
attentional vigilance (hypothesis 1), we propose this:
H2: A warning sign icon with more (vs. less) perceived
movement will result in faster reaction time.
Hypothesis 2 suggests that the desired behavior in a sign (e.g.,
stopping) will occur faster with a warning sign icon that conveys greater perceived movement. Furthermore, warning sign
icons able to evoke greater perceived movement should lead
to heightened levels of perceived risk. Based on these two
H3: A warning sign icon with more (vs. less) perceived
movement will result in earlier stopping behavior
(i.e., the stopping will occur farther back from the
An overview of our conceptual framework is depicted in figure
3. It is important to note that while perceived movement evoked
by icons may impact behavior across multiple domains, we
operationalize our research to icons that are within the domain
of warning signs, with a particular emphasis on traffic sign
icons, due to their important downstream behavioral consequences and their static visual nature.
Because the context of driving behavior is particularly
prone to demand effects stemming from learned compliant
behavior, we design our studies so that there is no correct
answer (e.g., the question Would you stop your car if you
saw a Stop sign? would be prone to demand effects since
there is only one correct answer). In addition, in two of our
studies, we design car driving scenarios that allow us to test
simulated driving behavior and reactions. We use multiple
methodologies and technologiestime to first fixation and
area of interest visits using an eye tracker (study 1), reaction
times measured in a driving simulation (study 2), and behavioral intentions captured through click-data heat maps
(studies 3A, 3B, and 4). Next we discuss our studies.


In study 1, we examine how perceived movement affects
attention. We have hypothesized that a higher dynamism
icon will attract earlier attention than a lower dynamism
icon (hypothesis 1a). We use eye tracking technology to
measure the time to first fixation on our area of interest,
which is the time in milliseconds from when the scene is
shown to a participant until the participants eyes fixate on
the traffic sign. We also hypothesized that a higher dynamism icon will evoke greater attentional vigilance than a

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

to react to; thus, our attention system evolved to detect animacy

automatically and quickly (New, Cosmides, and Tooby 2007),
with attentional prioritization given to animate objects before
the stimulus is fully identified (Ptak and Fellrath 2013). Evolutionary psychology further shows that the increased attentional vigilance triggered by animacy is domain specific, so
that such perceptual tunings work to foster survival (Nairne
et al. 2009). Higher perceived risk, or an increase in vigilance,
leads to greater detection of possible dangers. Consequently,
when scanning a scene containing potential risk, animacy gets
attentional priority and increases vigilance.
Given the considerable overlap in neural activity between actual movement and perceived movement (Kourtzi
and Kanwisher 2000), we predict that perceived movement
should have similar consequences on behavior as actual
movement. In this research, we examine the ability of dynamic iconography to create alertness and vigilance when
it comes to processing warning signs. Specifically, in an
alert state (e.g., when driving), icons that evoke greater perceived movement (dynamic imagery), being more vivid and
evoking more perceived risk, are likely to attract earlier
attention and lead to greater attentional vigilance within the
environment. This attentional vigilance should be exhibited
by participants shifting attention to the surrounding area
upon viewing a warning sign icon with more (vs. less) perceived movement. The warning sign icon, which will appear
to convey movement, will then be revisited, leading to additional shifting of attention back and forth between the
warning sign icon and the surrounding area. We operationalize this within our eye tracking study by examining the
number of visits to the areas surrounding, but not including,
the warning sign following initial exposure to the warning
sign icon.
Specifically, we predict that:




Stimuli and Pretests

We created eight road signsfour of lower dynamism
and four of higher dynamism. In designing the stimuli, we
tried to ensure that the lower and higher dynamism sign
icons contained the same information to minimize any other
visual confounds. All signs were manipulations (but not

perfect copies) of existing signs used in the United States.

The signs are shown in figure 4.
In a first pretest, 240 people from an online pool (administered on Amazon Mechanical Turk) were randomly
assigned to view only one of the eight signs in a betweensubjects design. Participants were asked to rate the sign
on its familiarity, visual appearance, visual complexity, and
informativeness. The familiarity scale was a 7-point scale
adapted from Clark (1970) and Dahl, Manchanda, and
Argo (2001). The sign visual appearance, visual complexity index, and informativeness scales were 9-point scales
adapted from Cian et al. (2014; see the appendix for details). Results show no significant difference (all ps 1 .1)
between the higher dynamism and lower dynamism versions
of each sign icon on our measures of familiarity, visual
appearance, visual complexity, and informativeness.
A second pretest was conducted to ensure that the higher
dynamism and lower dynamism sign icons differed on perceived movement. One hundred and sixteen undergraduate
students from the University of Michigan participated in the
pretest in exchange for course credit. Each participant was


Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

lower dynamism icon (hypothesis 1b). We measured attentional vigilance by the number of visits outside the traffic
sign after the initial visit to the traffic sign. Shifting ones
focus from the sign to the surroundings would count as one
visit. Subsequent visits to the surrounding areas would accumulate after refocusing on the traffic sign. Therefore, attentional vigilance is a measurement of darting ones eyes
across the scene and coming back to the sign. (Note that in
viewing a scene with a target object, one tends to fixate on
the target object at some time, and then one leaves it and
fixates on something else, and then reverts back and fixates
on the target object or fixates on another part of the scene;
this goes on repeatedly.)


Main Study
Fifty undergraduate students from Brigham Young University completed study 1 in exchange for course credit.
Since we are exploring the consequences of perceived movement within the context of traffic sign icons, in all studies
using traffic signs, we screened participants for only those
living in the United States and possessing a drivers license.
Data were collected using a high frequency (120 Hz) eye
tracker (Tobii T120) that was able to collect raw eye movement data points every 8.3 milliseconds. This eye tracker
is integrated into a 17-inch TFT monitor and has no visible
tracking devices that might affect participants behavior.
The eye tracker uses near infrared illumination to create
reflection patterns on the viewers cornea and pupil, with
two image sensors that capture images of the eyes.
Participants came individually into a conference room
every 10 minutes. The experimenter invited each participant
to take a seat in front of the eye tracker. Each participant
was taken through a calibration procedure before the eye
tracking recording was started. Upon failing calibration, participants chairs were adjusted, or the monitor was adjusted,
to provide accurate recognition of their eyes. Participants
were instructed that they were to evaluate several different
pictures of landscapes from the United States as if they were
driving on a road trip and that the scenes were shown from
a first-person perspective in a car. All instructions and stimuli
were presented on the 17-inch TFT monitor in full-color
bitmaps with a 1,280 # 1,024 pixel resolution.
Each scene was presented for 10 seconds and then advanced
on its own. Between each scene, participants saw a fixation
clue () in the middle of the screen for 1,000 milliseconds.

This was done to center fixation and thereby ensure that every
scene had the same centered attentional focus.
Participants were randomly assigned either to a condition
with only lower dynamism sign icons or to a condition with
only higher dynamism sign icons. In both conditions, the
participant evaluated all four signs in the same order (rock,
seesaw, snowmobile, horse).

Results and Discussion

Time to First Fixation. For measuring fixation, we created a specific area of interest around the traffic sign. An
identically-sized area of interest was applied to all scenes.
Our dependent measure was the time to first fixation in the
area of interest from stimulus onset. We ran a repeatedmeasures, mixed-model ANOVA with sign (rock/seesaw/
snowmobile/horse: within-subjects, repeated factor) and dynamism (lower/higher: between-subjects factor) as independent variables and time to first fixation as the dependent
variable. Figure 6, part a, shows the means for each scene.
The ANOVA shows a significant main effect for dynamism
(Mlower dynamism p 1.46, Mhigher dynamism p .82 seconds; F(1, 48) p
4.23, p ! .05) but not for its interaction with sign (F(3, 144)
p 1.66, p 1 .1). The results of the ANOVA and the mean
times to first fixation show that a higher dynamism sign decreases time to first fixation and does this similarly for all four
signs. This is consistent with our hypothesis (hypothesis 1a).
The ANOVA also reveals a significant main effect for sign
icon (F(3, 144) p 9.15, p ! .01), showing that some sign icons
(whether in lower or higher dynamism form) lead to faster
times to first fixation than others.
Attentional Vigilance. Per our hypotheses, dynamic imagery evokes greater attentional vigilance within the environment (hypothesis 1b). If people are generally more vigilant
after seeing a more dynamic sign icon, we should find a greater
scanning to other areas around it. We analyzed the eye tracking
data for number of visits to the areas surrounding, but not
including, the warning sign following initial exposure to the
warning sign icon (number of re-fixations).
We conducted a similar repeated-measures, mixed-model
ANOVA with number of re-fixations as the dependent variable
(see fig. 6, part b, for the means). The ANOVA shows that
higher dynamism sign icons led to significantly more visits
outside the sign than lower dynamism sign icons (Mlower dynamism
p 2.04 visits, Mhigher dynamism p 2.53 visits; F(1, 48) p 4.39, p
! .05), suggesting greater attentional vigilance when movement
is anticipated. Neither the main effect for sign (F(3, 144) p
1.71, p 1 .1) nor the interaction between sign and dynamism
(F(3, 144) p .72, p 1 .5) were significant.
The eye tracker results in study 1 support our hypotheses
(hypotheses 1a and 1b) that, in a situation of alert (imagining
driving, in this case), static pictures evoking more (vs. less)
perceived movement are able to draw attention more quickly,
resulting in an earlier fixation. Moreover, dynamic imagery
increases attentional vigilance within the environment, as indicated by a higher number of visits beyond the sign.

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

randomly assigned to view either the lower or the higher

dynamism version of each sign icon. We used a two-item
scale to measure perceived movement adapted from prior
literature: How much movement did you see in the icon
depicted in the traffic sign? (1 p No movement at all, 9
p A lot of movement); and How dynamic was the icon
depicted in the traffic sign? (1 p Not at all dynamic, 9 p
Extremely dynamic; r p .78, p p .01; Cian et al. 2014).
As expected, we found a significant difference between the
higher dynamism and lower dynamism version of each sign
icon on our measure of perceived movement (all ps ! .05),
with the higher dynamism sign icon leading to more perceived
movement than the lower dynamism sign icon.
After the pretest, we inserted these eight signs in realistic
road settings. Specifically, we created two sets of four stimuli (eight total stimuli) depicting a first-person driving view,
with four lower dynamism and four higher dynamism sign
icons (fig. 5). We used four different driving scenes and
backdrops. Each scene contained a road from a first-person
perspective, as well as the windshield and hood of a car,
which did not change across scenes. In each scene, we placed
a traffic sign (lower or higher dynamism) at a different location on the right side of the road to avoid participants
fixating in the same exact area across scenes.




Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016


While study 1 tests for attention, study 2 tests for quickness of response to a traffic sign icon. Specifically, we had
participants watch a video driving simulation from a drivers

perspective. To the right of the driving scene, we placed

traffic signs. We were interested in how quickly participants
would respond to the signs in this scenario. The drivers
had to press a key on the keyboard to indicate their reaction.
If, as we hypothesize, the body sees a sign and gets ready
for action in the active concept, then even though pressing
a key is not the same as pressing the brakes or turning the



In order to design a realistic and involving reaction time
study, we created a driving video simulation from a firstperson (drivers) perspective. We edited a video showing a
drive along a country road as seen from the drivers point
of view. The video was 16 seconds long.
We were interested in participants deliberate reactions to
traffic warning signs and not participants reaction time to
anything appearing on the screen (resembling a mindless
reaction time to a visual appearance). Thus, if we were to
have only traffic warning signs appear on the screen and
have participants hit a key when they appeared, participants
would hit a key when anything appeared on the screen
(since everything appearing would be a traffic sign). In order
to avoid this problem and still keep the driving simulation
realistic, we created two types of traffic signswarning
signs and informative signs. These two signs warranted
different reactions from the participants, hence necessitating
some deliberation and not mindless hitting of a key. We instructed participants that warning signs signaled unexpected
conditions on or adjacent to a highway, street, or private road
and to situations that might not be readily apparent to road
users and that these may ask for a reduction of speed. They
correspond to the warning signs category in the American
drivers manuals. We also told participants that informative
signs were signs used to identify a direction or a general
service. Informative signs provide additional information but
do not require a reduction of speed. They correspond to the
general information/general service signs category in the
American drivers manuals (we adapted these definitions from
the 2012 DOT Manual on Uniform Traffic Control Devices).

We asked participants to press the w key for a warning sign

and the i key for an informative sign.
In all, we created 12 road signs4 informative signs and 8
warning signs. For warning signs, we used the same 8 signs
created in study 1, which were variants of existing signs (4
with lower dynamism and 4 with higher dynamism). The 4
informative signs are shown in figure 7 (these signs were
adapted from the 2012 DOT Manual on Uniform Traffic Control Devices general information/ general service signs category). For the informative signs, we use the same background
(yellow diamond) as that in the warning signs, so that participants have to make their decision based on the icon contained
in the sign and not on the shape of the traffic sign.
Note that reaction time studies are typically based on
pictures or words (Gawronski and Payne 2010; Fiske, Gilbert, and Lindzey 2010). In our reaction time study, we use
a first-person driving video simulation, necessitating a synchronization of video and reaction time measurement, something that was far from trivial. Additionally, since reaction
times vary widely across individuals (Broggi et al. 2008;
Manning, Tolhurst, and Bawa 2012), we wanted to look at
within subject differences in reaction times for lower versus
higher dynamism warning sign icons. However, we did not
want to give participants lower and higher dynamism versions of the same warning sign icons as this may appear
suspicious and lead to possible demand effects. Additionally,
participants may become faster at recognizing a sign because
of increased familiarity (seeing two versions of it).
To deal with the concerns mentioned above, we programmed the video so that the 16-second video looped four
times on the left half of a 14.5-inch computer screen. On
the right half of the screen, one warning sign and one informative sign appeared (in random order) for 1 second each.
The first sign (informative or warning) was programmed to
appear after 5 seconds, the second (warning or informative,
respectively) after 8 seconds. Thus, each participant saw a
64-second (16 seconds # 4) video and was exposed to all

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

wheel, the resulting action should be faster when viewing

a higher (vs. lower) dynamism sign icon. However, given
the difference between the actions, this is a conservative
test of our theory for faster driver reaction time to more
versus less dynamic traffic sign icons (hypothesis 2).



A pretest was conducted to ensure that people were able to
discern informative from warning signs. Sixty undergraduate
students from the University of Michigan participated in the
pretest in exchange for course credit. We presented the participants the definitions of warning and informative signs and then
showed them two examples of each. The examples used differed from the signs participants were tested on. Participants
were told that they would be viewing several traffic signs and
that they were to determine if the sign was a warning sign or
an informative sign, a binary choice. Furthermore, they were
instructed to base their decision on the icon depicted on the
sign and not on the shape of the traffic sign (yellow diamond),
which was the same for all signs.
Each participant next saw the whole battery of 12 signs (4
informative signs, 4 lower dynamism warning sign icons, 4
higher dynamism warning sign icons) in random order. They
were allowed to view each sign for as long as they wished and
were asked to indicate whether each sign was an informative
sign or a warning sign. The error rate was small and acceptable
(7.50%); therefore, participants recognized the warning (informative) signs as such. To further analyze the data, we used a
generalized estimating equations (GEE) model, which is a generalized form of logistic regression for choices observed under
a within-subjects design (Ge, Haubl, and Elrod 2012; Liang
and Zeger 1986). Within this model, the 12 signs were used
as the independent variable and choice (correct/incorrect) as
the dependent variable. Neither the main effect of sign (Wald
p 6.53, 11 df, p 1 .8) nor the pairwise comparisons of all sign
combinations were significant (ps 1 .1), indicating that participants were able to discern if a sign was a warning or an
informative sign without any significant difference in terms of
the likelihood of across the 12 signs.

Main Study
Procedure. Two hundred and seventy-five undergraduate students from the University of Michigan participated
in the main study in exchange for course credit. First, participants read the test instructions. They were told that this
was a reaction time driving test, and they were instructed
to imagine themselves driving while answering as quickly
and as accurately as possible. Similar to the pretest, definitions and examples of warning signs and informative signs
were provided. Participants were asked to answer based on
the icon depicted on the sign and not on the shape of the
traffic sign itself. They were instructed to press the i key
when they saw an informative sign and the w key when
they saw a warning sign. Second, participants performed a
practice trial (in which we used four signs different from
the main study). Inaccurate categorizations were followed
by the word incorrect in red font for 1.5 seconds. Finally,
participants went on to the main study. Reaction times were
captured after the onset of each sign. Figure 8 shows a
screenshot taken during the study.

Results and Discussion

We followed the data analysis procedure outlined by Meier
et al. (2007), Ratcliff (1993), and Robinson (2007) for analyzing
reaction time data. First, we deleted inaccurate trials where the
informative (warning) sign was misread by the participant as
a warning (informative) sign (12.27%). Second, we log-transformed the response times for correct trials to reduce the typical
skewed distribution of reaction time data. Finally, we replaced
log-latencies faster or slower than 2.5 standard deviations from
the log-latency mean with these cutoff scores. Log transformation followed by 2.5 standard deviations replacements seems
to be one of the best procedures to prevent capitalizing on
chance (Robinson 2007). Analyses were conducted on logtransformed data, but means are also reported in raw milliseconds to facilitate comprehension. The reaction time means for
each sign are reported in figure 9.
Note that what we are interested in is the difference in
reaction times between lower and higher dynamism traffic
sign icons. The reaction needed for the two signs is the same
in that they require pressing the same key (w with the left

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

four informative signs and all four warning signs. Two of

the four warning sign icons were in their lower dynamism
form and two were in the higher dynamism form (as discussed above). Whether the sign icons appeared in their
lower or higher dynamism version was randomly assigned
across subjects. As stated earlier, on seeing a traffic sign,
participants had to press the w key for warning sign or
the i key for informative sign.




NOTE.Screenshot of what participants saw during the reaction time study. The 16-second video appears on the left; a sign appearing for 1
second is on the right.

Supporting our hypothesis 2, a repeated measures ANOVA

showed that sign type had a main effect on reaction time
(Wilkss Lambda p .97, F(2, 256)p 4.67, p ! .05). Planned
follow-up contrasts revealed that participants reacted significantly faster to warning signs with higher dynamism icons than
to warning signs with lower dynamism icons (F(1, 257) p
6.23, p ! .05) or to informative signs (F(1, 257) p 8.41, p !
.01). There was no significant difference between warning signs
with lower dynamism icons and informative signs (F(1, 257)
p .02, p 1 .8). Similar results were obtained if we omitted the
informative signs and just performed our analyses using the
lower and higher dynamism sign icons.
Study 2 exhibits additional behavioral consequences of
greater perceived movement within traffic signs. Specifically, higher dynamism sign icons led to quicker reaction
time than lower dynamism sign icons or informative signs
(as predicted in hypothesis 2). In addition, study 2 increases
the external validity of the findings by placing them within
a more realistic driving simulation. Note also that a difference of 50 milliseconds, in a car traveling at 60 miles per
hour, is a distance of 4.4 feet and can mean the difference
between having an accident versus not.


Studies 1 and 2 test attention and quickness of response to
a traffic sign icon, respectively. Study 3 tests whether more
(vs. less) perceived movement within traffic warning sign icons
leads to more compliant and cautious behavior, as indicated by
where one decides to stop a car. We hypothesize that drivers
will stop farther back from the traffic sign when perceived
movement resulting from the sign icon is higher (i.e., when
the observer is more likely continuing the movement in her
mind) than when it is lower (hypothesis 3). We test this in two
progressive studies in different consumer contexts.

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

hand) and the only thing that varies is the dynamic nature
of the icon in the sign seen. For the informative sign, participants pressed the i key with their right hand, adding
a hand confound, but with most participants being righthanded this would make the informative signs a conservative
control. Remember that each participant saw eight traffic
signs (four informative, two higher dynamism, and two
lower dynamism) and that whether the participants saw the
lower or higher dynamism version of the warning signs was
randomized across subjects.
We performed two types of analysesone was a regression analysis on individual sign data and one was a repeated
measures analysis on data pooled across signs. We begin
with the regression. We are interested in the difference in
reaction time between the lower and higher dynamism sign
icons. A regression with effect coded dummies for the four
warning signs (sign type), an effect coded dummy to indicate
lower and higher dynamism (dynamism), and their interactions, yielded significant effects for dynamism (b p .09,
t p 2.84, p ! .01). No interactions were significant ( ps
1 .4). The significant effect for dynamism supports our hypothesis 2 that a higher dynamism sign will evoke faster
reaction time. The nonsignificant interaction between sign
type and dynamism indicates that the difference in reaction
times for a particular lower dynamismhigher dynamism
pair of warning sign icons is not significantly different from
the other three warning sign icons. Since we are specifically
interested in the difference in reaction times between lower
and higher dynamism warning sign icons, the lack of the
interaction allows us to pool data across warning sign icons.
In a repeated measures analysis, we pooled the data across
signs for each of three sets of signs that each subject saw
thus, we had three pooled reaction times per participant, for
the informative signs, higher dynamism sign icons, and lower
dynamism sign icons (the regression analysis discussed previously also justifies the pooling). The reaction time means for
each of these three variables are reported in figure 10.




We created four stimuli, with each stimulus containing a

road with a road sign of children crossing as seen from above.
All stimuli (800 # 600 pixel images) were identical except
for the icon depicted in the warning sign. Figure 11 shows an
example of the stimulus, with all four signs shown below for
illustrative purposes. Two of the signs had visual icons of children (with lower and higher dynamism), and these served as
the main experimental stimuli. They are manipulations (but not
perfect copies) of the existing sign used in the United States
to represent a school crossing. We also had two control conditions that used verbal signs (children ahead and running
children). These verbal depictions allow us to test whether
differences in behavior are due to differences in inferred conceptual meaning from the signs or if dynamic imagery from
visual cues evokes unique behavioral reactions. Thus, we had
a one-way design with four conditions (lower dynamism visual,
higher dynamism visual, verbal controlchildren ahead, and
verbal controlrunning children).
Pretests. Similar to study 1, in a first pretest conducted
online (n p 55), participants rated the two visual traffic
signs on dimensions of familiarity, visual appearance, visual

complexity, and informativeness (all ps 1 .2). A second

pretest with 110 undergraduate students from the University
of Michigan revealed significant differences between the two
visual signs on perceived movement ( p ! .05) in the expected direction.
Main Study. One hundred and fourteen participants from
an online panel participated in the study in exchange for money.
Participants were randomly assigned to view one of the four
conditions and were instructed to imagine themselves driving
on the road depicted in the picture. They were further told to
click on the location of the road where they would start to slow
down their vehicle in response to the traffic sign. Using
JavaScript, we programmed the survey to place a small car
where the participant clicked. The final position of the car
represented our main dependent variable of interest.
Results and Discussion. Our main dependent variable of
interest (car position) was calculated in terms of geometrical distance from the origin (bottom left of the image)
to the point where participants clicked with the mouse.
Lower numbers represent shorter distances from the origin
and consequently longer distances from the sign.
We conducted a one-way ANOVA with the sign condition

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

Study 3A: Children Crossing



Study 3B: Shopper Crossing

Study 3B tests if the results of study 3A carry over to
another consumer setting and with completely unfamiliar
signs. In recent years, 30%40% of the pedestrian injuries
occur within parking lots (NiTS 2012; Tschida 2013), mostly
at shopping malls or grocery stores. We created an aerial
view (440 # 550 pixels) of a mall parking lot with a road
sign. The road sign icon portrayed a shopper with a shopping
cart. We made two such icons, one having more perceived

movement than the other. We manipulated the image of the

shopper and shopping cart to ensure that the two icons contained identical content. The only difference between the
stimuli was how the images were stylized (lower/higher dynamism; fig. 13). Neither version of the sign corresponded
to an existing traffic sign in the United States.
Pretests. As in the previous studies, a first pretest with
54 participants from an online pool revealed no significant
difference between the two sign icons (all ps 1 .2) on familiarity, visual appearance, visual complexity, and informativeness. A second pretest conducted with 40 participants
from an online pool revealed significant difference between
the two sign icons on perceived movement ( p ! .01).
Main Study. Ninety participants from an online panel
participated in the study in exchange for money. Participants
were randomly assigned to view one of the two maps and
were instructed to imagine themselves driving along the blue
arrow (black arrow if the reader is seeing the figures in
black and white) through the parking lot depicted in the
picture. Participants then clicked on the map where they
would stop their car. In place of their cursor, we programmed
a car to appear. The rest of the instructions and the procedure
were similar to study 3A.
Results and Discussion. We conducted a one-way
ANOVA with icon dynamism (lower/higher) as the independent variable and car position as the dependent variable. As in the prior study, participants indicated earlier
positions to begin slowing down when shown a higher dy-

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

as the independent variable. The resulting ANOVA was significant, showing a difference in car position among the signs
(F(3, 110) p 3.10, p ! .05). Simple-contrasts provide support
for our hypothesized effects (see fig. 12): the means indicate
that participants would begin to slow down earlier when
shown a higher dynamism (vs. a lower dynamism) visual sign
(F(1, 110) p 4.69, p ! .05). Additionally, they would slow
down significantly earlier with a higher dynamism visual sign
versus both the verbal control children ahead sign (F(1,
110) p 7.08, p ! .01) and the verbal control running children sign (F(1, 110) p 6.54, p ! .05). There was no significant difference between the lower dynamism sign and the
two verbal control signs (all ps 1 .6). Figure 12 gives the
heat map for each condition. Each heat map is a graphical
representation of data displaying where participants placed
their car. Together this suggests that it is the dynamic imagery
evoked from visual cues that affects behavior and not simply
the activation of the concept of movement.



namism versus lower dynamism sign icon (Mlower dynamism p

377.79 pixels, Mhigher dynamism p 352.31 pixels; F(1, 88) p
4.57, p ! .05). Figure 14 depicts a heat-map summary of
the click data.
Study 3B replicates the results of study 3A, while generalizing the findings in a different setting with novel icons.
The two studies together support our hypothesis (hypothesis
3) showing the effect of icon dynamism on behavior.




NOTE.Above, a screenshot of study 3A showing the location of the sign inside the map and the car that appeared when subjects clicked
with their computer mouse. Below, the four signs used in study 3A.

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

The purpose of the study 4 is twofold. First, we seek

further support for hypothesis 3 and formally test our proposed conceptual framework. Per our conceptual framework
(fig. 3), we suggest that higher dynamism icons lead to
greater mental completion of the depicted movement than
lower dynamism icons. This imagined movement (e.g., a
rock falling on the road) is represented by a more vivid

image of the depicted act (Callow et al. 2006; Smallwood

et al. 2004).
The second purpose of study 4 is to explore if the underlying
process for the effect of icon dynamism stems from the dynamic
imagery the icon evokes or if it requires participants to engage
in directed mental simulation. Mental simulation is defined as
the construction of hypothetical scenarios in the form of imagined situations (Taylor and Schneider 1989). In studies 13,
we asked participants to imagine themselves driving through
the scene depicted in the picture (i.e., directed mental simulation). It is possible that when participants are asked to imagine
themselves in the scenario, they are also more likely to imagine
the completion of the act shown in the warning sign. While
we propose imagery to be the underlying mechanism behind
how the icon impacts the completion of movement, we argue
that the imagery is contained within the icon itself. In other
words, when one sees an image conveying motion, one completes the motion, with or without directed mental simulation.
Thus, in study 4, we instruct participants either to imagine




Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016




Stimuli and Pretests

In study 4, we operationalize dynamic imagery using a
warning sign (wet floor) inside a grocery store aisle. This
set of stimuli allows us to keep our focus on warning signs,
while changing the consumer setting. Similar to studies 3A
and 3B, we created an aerial view (417 # 550 pixels) of
the grocery store aisle with the sign placed near the top of
the image. The sign contained an icon of a man slipping on
the floor. Both the higher and lower dynamism icons were
identical in form, with the only difference being the angle
of rotation of the man depicted in the icon. The lower dynamism icon depicted a man more upright, with the higher
dynamism icon rotated 45 degrees counterclockwise (see
fig. 15).
We conducted a pretest with 40 students from the University of Michigan. Each participant was randomly assigned to view one of the two pictures shown in figure 15.
Participants were told that they would be viewing and evaluating a sign shown in a grocery store aisle. After viewing
the picture, we asked participants to evaluate the sign on

dimensions of familiarity, visual appearance, visual complexity, and informativeness, as used in the prior studies (all
ps 1 .1). Furthermore, with an open-ended question, we
asked participants to describe the meaning of the sign. Two
independent trained coders, who were blind to the hypotheses, coded the responses. For the higher dynamism icon,
20 out of 20 participants correctly recognized the sign. For
the lower dynamism icon, 17 out of 20 participants correctly
recognized the sign. In a second pretest, 48 people from an
online pool revealed significant differences between the two
signs on perceived movement ( p ! .05).

Main Study
Design. Study 4 was a 2 (directed mental simulation:
yes, no) # 2 (icon dynamism: lower, higher) between-subjects design. Our dependent variables were stop position,
perceived risk in the scenario, imagery vividness, and sign
Procedure and Measure. One hundred and forty-three
participants from an online pool completed the study in
exchange for money. Each participant was told that s/he
would be evaluating a shopping scenario. To induce an alert
state, we began the questionnaire instructing participants that
they would be viewing warning signs within a grocery store
setting. Following these initial instructions, participants
were randomly assigned to directed mental simulation conditions. In the yes directed mental simulation condition,
participants were given the following instructions: We ask
that you use your imagination to make your responses. Visualize yourself in the shopping scenario and use the power
of your imagination to make your choice. In the no directed
mental simulation condition, participants were given the following instructions: We ask that you be well-reasoned in
your responses. Do not use your imagination when making
your decisions. Rather, try to make a logical choice. We
could have chosen to not give any instructions to participants

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

themselves in the scene (directed mental simulation, replicating the instructions of study 3A and study 3B) or to not
imagine themselves in the scene (no directed mental simulation). We predict that even in the no directed mental
simulation condition, icon dynamism will impact participants behavioral response through our proposed process.
Consequently, we propose that in both conditions (directed
mental simulationyes, no) icon dynamism will have a significant impact on participants behavioral responses and
that this will occur through movement completion (captured
by imagery vividness). Thus, we predict that icon dynamism
will have a main effect on behavior mediated through imagery vividness (as shown is fig. 3) irrespective of whether
directed mental simulation is present or absent. The study
is therefore designed to identify the source of imagery underlying the effects.



lowing dimensions: 1 (Not at all vivid/ intense/lifelike/sharp/

defined); 9 (Extremely vivid/intense/lifelike/sharp/defined);
a p .96; adapted from Bone and Ellen 1992). Finally, sign
familiarity was measured using the same items used in the
previous studies (a p .89).

Results and Discussion

We first conducted a 2 # 2 ANOVA on our dependent
variables with directed mental simulation and icon dynamism as our independent variables. As predicted, the interaction of directed mental simulation and icon dynamism was
not significant for stop position ( p 1 .5), perceived risk ( p
1 .3), or imagery vividness evoked by the icon ( p 1 .9).
Additionally, the main effect of directed mental simulation
was not significant for stop position ( p 1 .2), perceived risk
( p 1 .1), or imagery vividness ( p 1 .8) either. Thus, for our
primary analysis, we collapsed across the two directed mental simulation conditions.
We next conducted an ANOVA with dynamism (lower/
higher) as the independent variable and stop position as the
dependent variable. As in the prior studies, participants indicated earlier positions to begin slowing down (indicated
by lower pixel numbers along the y-axis) when shown a
higher dynamism versus lower dynamism icon (Mlower dynamism
p 320.23 pixels, Mhigher dynamism p 284.47 pixels; F(1, 141)
p 4.18, p ! .05). An ANOVA with icon dynamism as the
independent variable and perceived risk as the dependent variable revealed a similar pattern of results on z-scores (ranging
from 2.87 to 1.54). Specifically, the higher dynamism icon
led to significantly higher perceived risk than the lower dy-

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

in the no directed mental simulation condition. However, as

McGill and Anand (1989) note, people tend to imagine
themselves in a situation even in the absence of instructions
to imagine. Therefore, we adapted the logic instructions
from McGill and Anand (1989) to discourage the use of
mental simulation. To ensure participants correctly read the
manipulation, each was asked to indicate their understanding: I understand that I will visualize myself in the scenario
and use the power of my imagination to make my choice
or I understand that I will try to make the most logical and
rational choice, depending on the condition. Participants
were asked to answer yes or no. Only participants who
answered yes continued the survey.
Next, participants were randomly assigned to either the
lower or higher dynamism icon conditions. They were told
to click on the location in the aisle where they would start
to slow down in response to the sign. The final click position,
indicating where the participant would start to slow down
(stop position), was our first dependent variable of interest.
Participants were then asked for their perceived risk in the
scenario, imagery vividness evoked by the icon, and sign
familiarity. Perceived risk was measured with two items
(How cautiously would you walk down this grocery store
aisle? anchored on 1 (not at all cautiously) to 9 (very cautiously), and Relative to your normal pace, how slowly
would you walk down this grocery store aisle? anchored
on 1 (the same speed as normal) to 100 (extremely slower
than normal); r p .73, p ! .01 on the z-scores). Imagery
vividness evoked by the icon was measured by asking participants to rate the mental image they formed of themselves
slipping on the floor while looking at the sign (on the fol-



namism icon (Mlower dynamism p .20, Mhigher dynamism

p .19; F(1, 141) p 6.94, p ! .01). Our third ANOVA examined the impact of dynamism on vividness evoked by the
icon. The higher dynamism icon led to greater vividness of
the action depicted in the icon than the lower dynamism
icon (Mlower dynamism p 5.03, Mhigher dynamism p 6.04; F(1, 141)
p 8.90, p ! .01). A final ANOVA showed that the two
icons did not differ in term of familiarity (Mlower dynamism p
2.37, Mhigher dynamism p 2.44; F(1, 141) p .114, p 1 .7).

Our findings show the behavioral consequences of perceived movement from static visuals. Across five studies,
using multiple methodologies and technologies, a higher
dynamism (vs. lower dynamism) icon results in earlier attention to, more attentional vigilance, and faster reaction to
signs (studies 1 and 2) and safer behavior (studies 3A, 3B,
and 4). Specifically, using eye tracking in study 1, we show

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

Serial Multiple Mediation. With significance across our

dependent measures between conditions of icon dynamism,
we tested the process underlying the results. Specifically,
we tested the path from icon dynamism (lower/ higher) r
imagery vividness (as result of mental completion of movement) r perceived risk r behavior (stop position) using a
multiple mediator model with serial mediation and bootstrapping (5,000 resamples; Hayes 2013, model 6). The examination of the serial indirect effect of icon dynamism on
stop position through both mediators (vividness first, and
perceived risk second) shows that the path from icon dynamism to vividness is significant (B p 1.01, p ! .01), as
is the path from vividness to perceived risk (B p .10, p !
.05), and also from perceived risk to stop position (B p
41.07, p ! .01). This full path of serial indirect effects is
significant, with a 95% confidence interval between .83 and
11.29. In addition, when both mediators are included in the
model, the direct effect of icon dynamism on stop position
is no longer significant ( p p .1).
Thus, we find support for our hypothesized model that
higher dynamism within the icon is represented by a more
vivid image of the act depicted, affecting perceived risk,
and ultimately behavior. Despite the usage of our manipulation of directed mental simulation in prior research (McGill and Anand 1989), it is possible that in our no directed
mental simulation condition the instructions to suppress imagery may have still led to imagery, as spontaneous image
generation has been shown to be a rather ubiquitous process
(Jiang et al. 2014). Indeed, attempts to control ones thought
process may lead to ironic effects of intention to control or
concentration, with imagery generation resulting from instructions not to imagine (Wenzlaff and Wegner 2000).
However, our results suggest that the process is driven by
the spontaneous generation of imagery through dynamic
iconography and not from directed mental simulation. Instructing participants to imagine themselves in the scene
does not make them more likely to imagine the completion
of the act shown in the warning sign.

that higher dynamism icons draw earlier attention and increase attentional vigilance (hypothesis 1) when compared
with lower dynamism icons. In study 2, we use a driving
simulation video and show that higher (vs. lower) dynamism
icons lead to quicker reaction time (hypothesis 2). In studies
3A and 3B, increased dynamism within a static visual of a
traffic sign leads to stopping at a safer distance from a potentially dangerous situation (hypothesis 3). Additionally, in
study 4, we provide support for the conceptual framework
we propose (see fig. 3) and exhibit the process underlying
our effects.
Our work contributes to the literature on dynamic imagery
(e.g., Cian et al. 2014; Clark and Paivio 1991) by exploring
the behavioral consequences and the underlying process of
such imagery. Our work also contributes to the literature
exploring the impact of visual cues on consumer behavior
(e.g., Greenleaf and Raghubir 2008; Patrick and Peracchio
2010; Wang and Peracchio 2007) by showing how a variation in the dynamism of a static picture can affect consumer
attention, risk perception, and behavior. Our exposition of
the effects of perceived movement on behavior provides a
foundational framework for future research within consumer
psychology and behavior. However, further exploration of
when and how dynamic imagery is elicited, as well as how
it differs fundamentally from static imagery, is important.
The implications stemming from dynamic imagery also
need further exploration. For example, do the cognitive processes that underlie dynamic imagery require additional resources relative to static imagery? A core consequence of
perceived movement is misplacement of the object depicted
in memory in the direction of the movement (Freyd and
Finke 1984); however, should there be additional resources
devoted to form a dynamic image, there may be positive
memory consequences, such as increased memory for the
stimulus. From both a consumer behavior and cognitive psychology perspective, understanding how the brain interprets
dynamic visual cues, re-creates them within imagery, and
ultimately remembers them provides a compelling area for
future research.
Across our studies we find persuasive support for our
conceptual model. However, we do not test the entire model
in one study, but rather use a chain of studies to provide
support for the underlying process (see Lee and Schwarz
2012; Spencer, Zanna, and Fong 2005 for similar methodology). As we were focused on identifying and exploring
the basic elements of the process, we did not fully explore
the exact order between the elements, nor their boundary
conditions. For example, in addition to the warning signs
we used in our studies, we also used informative signs. It
is possible that such informative signs, even when imbued
with higher dynamism, follow a different path in affecting
One possible limitation of our current set of studies is the
manipulation of an alert state prior to stimulus presentation.
Being in an alert status may indeed influence attention. Human attention has a very high degree of flexibility; it can
be directed on small and specific regions or it can have a


icons may be driven by expectations. The interplay between

top-down and bottom-up processes provides an intriguing
area for future research.
In all our studies, we used frozen motion as a way to
evoke a perception of movement from static icons in traffic
signs. However, other antecedents of perceived movement
could also be explored and tested for their efficacy. Looking
at the cognitive psychology literature, visual friction, a concept wherein visual cues touching each other inhibit their
perceived movement (Hubbard 1995; Kerzel 2002), could
be used to alter dynamism within a consumer context.
Within traffic signs specifically, drawing borders around the
existing icons may inhibit perceived movement, ultimately
reducing desired behavior. For example, a deer crossing sign
with a thick black border may convey less dynamism than
one without the border, leading to slower reaction times.
Drawing from the art literature, we can also find other methods of conveying movement, such as color (Mazow 2013)
and repetition (Boccioni et al. 1910).
As the findings from our research have direct implications
on consumer welfare, our work also adds to the domain of
transformative consumer research and its primary focus on
individual and social well-being (Mari 2008; Mick 2008;
Mick et al. 2011). Since variations in traffic iconography
systematically resulted in safer behavior, this research could
impact accident-related injuries and even mortality rates
from traffic accidents. Secondary effects would be seen with
reductions in automobile repair expenses, as well as car
insurance health care costs. It is important to note that studying perceived movement from static images is particularly
relevant in a consumer setting. In most instances where icons
are used (e.g., traffic signs, warning icons, and informative
icons), these icons cannot be animated due to economic and
technical reasons. Furthermore, the icons often cannot be
accompanied by verbal descriptors for space limitations or
because of multiple languages being used in a country. Consequently, our research contributes to the understanding of
a basic characteristic of visual grammar, showing how perceived movement can affect cognitions and behavior (such
as time to first fixation, reaction times, risk perception, and
behavior) that are crucial in dangerous contexts. More specifically, these findings can be used to facilitate more responsible consumption.
While we have chosen to operationalize our research
primarily within the context of traffic icons, the practical
applications of our theoretical contributions extend well
beyond this domain. Since higher dynamism leads to behavioral reactions, then increasing dynamism in recycling
icons, consumption cues for health foods, or other domains
should also have behavioral implications. Exploration of
these additional contexts should prove valuable from public policy and consumer welfare perspectives. Our studies
show dynamic imagery to be an important and underexplored construct within both cognitive psychology and
consumer behavior. We hope that this work will spur further exploration of this topic.

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

more holistic and distributed focus (Pashler 1998). Neurological evidence has concluded that attention focuses only
on the location (or object) with the highest priority, and that
the locus of attention reflects converging sensory and goalrelated signals (Gottlieb, Kusunoki, and Goldberg 2005;
Koch and Ullman 1987; Ptak and Fellrath 2013). Thus,
based on our theoretical framework, the state of alert may
speed the initial attention directed to the warning sign because the detection of possible dangers represents a priority.
In absence of a state of alert, the ability of dynamic icons
to attract earlier attention may decrease, because our attentional priority is not focused on detecting possible threats.
However, once the warning sign is detected, this should
increase the state of alert, as well as perceptions of risk.
Because a dynamic icon leads to a spontaneous completion
of movement (independent from instructions), once the sign
is attended to, we predict the effects of dynamic iconography
to obtain as shown across our series of studies. As we do
not have empirical support for this claim, however, this question warrants future research.
The icons that we used in our studies (with the exception
of the shopping cart icons) were similar to those used in
the real world. It is possible that the level of familiarity with
the signs can affect our results. When viewing completely
unfamiliar visual stimuli, processing may, for example, be
limited to interpreting the base form of the stimulus and not
the movement of the stimulus. While we have pretested all
stimuli to be equated on novelty, it is possible that the level
of familiarity with the sign, or perhaps even the context the
sign is placed in, would moderate our effects.
In addition to the level of familiarity with the sign, the
type of behaviors depicted by the sign may evoke a form
of mimicry. While in studies 13 the signs referred to behaviors of external agents (e.g., rocks falling, children
crossing, horses running), the sign in study 4 referred to
ones own behavior (i.e., slipping and falling). The dynamism contained within the sign led to more cautious behavior, but another way to evoke the desired behavior is to
portray the icon in a way that leads to mimicry (e.g., a man
cautiously walking). In study 4, the two icons that we used
differed in terms of perceived movement but also possibly
in terms of mimicry. In the lower dynamism condition, the
sign could have been interpreted as a tiptoeing man, while
the higher dynamism condition represented a slipping man.
The first condition, while lower in dynamism, facilitates
people to mimic that action and walk more carefully. The
fact that the more dynamic condition led to higher vividness,
perception of risk, and preventive behavior allows us to
make a distinction between icons that evoke mimicry versus
attention/vigilance. However, this distinction needs further
study. In certain contexts, dynamism and mimicry may be
opposing forces. For example, based on our findings one
would want the children crossing sign to convey dynamism,
with running children being better than walking children for
driver behavior; however, for the pedestrians, mimicking
running across the street may have unintended consequences. Therefore, the resulting behaviors from the sign





The third author supervised the collection of data for
study 1 (eye tracking) by research assistants at the Marriott
School of Management, Brigham Young University, during
January and February 2014. The first author collected the
data for the other studies, from summer 2013 to May 2014.
The first author collected the data either at the Behavioral
Lab of Ross School of Business (University of Michigan)
or using the Amazon Mechanical Turk panel (the details are
described in the methods section of the studies). The first
author analyzed the data collected in all studies.


AAC (Advocate Auto Consultants) (2010), Parking Lots Account for 1.26 Million Car Accidents in the U.S. Annually,
with a 20% Jump, http://www.reuters.com/article/2010/12/
Adaval, Rashmi, and Robert S. Wyer Jr. (1998), The Role of
Narratives in Consumer Information Processing, Journal of
Consumer Psychology, 7 (3), 20745.
Anderson, Tony, Christine Howe, and Andrew Tolmie (1996), Interaction and Mental Models of Physics Phenomena: Evidence
from Dialogues between Learners, in Mental Models in Cog-

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

The familiarity scale was an adaption from Dahl et al.

(2001) and Clark (1970) and consisted of three items: How
often have you seen the sign before? (1 p I have seen
this sign very often, 7 p I have never seen this sign);
This is an existing sign (1 p Strongly agree, 7 p
Strongly disagree); and This is a well known sign (1 p
Strongly agree, 7 p Strongly disagree).
The sign visual appearance scale, the sign visual complexity index, and sign informativeness were adapted from
Cian et al. (2014).
The sign visual appearance scale consisted of three items:
I like the way the sign looks (1p Strongly disagree, 9
p Strongly agree); The sign is attractive (1 p
Strongly disagree, 9 p Strongly agree); and The sign
is aesthetically appealing (1 p Strongly disagree, 9 p
Strongly agree).
The sign visual complexity index was measured through
two items. In the first item, participants were instructed to
rate the complexity of the sign on a 9-point scale in which
1 indicated very simple and 9 indicated very complex. Complexity was defined as the amount of detail or intricacy of
line in the sign. Participants were told to rate the complexity
of the drawing itself rather than the complexity of the reallife object it represented. The second item consisted in asking: How would you evaluate this sign? (1 p Not complicated, 9 p Very complicated). Sign informativeness
was measured by asking participants to evaluate how informative was the sign, on a 9-point scale in which 1 indicated Not informative at all and 9 indicated Very informative.

nitive Science: Essays in Honour of Phil Johnson-Laird, ed.

Jane Oakhill and Alan Garnham, East Sussex, UK: Psychology, 24773.
Arbitron, Edison Media Research (2013), The National In-Car
Study, http://www.vehiclewrappricing.com/LiteratureRetrieve
ASIRT (Association for Safe International Road Travel)
(2014), 2013 Road Crash Statistics, http://www.asirt.org
Barrett, H. Clark (2005), Adaptations to Predators and Prey, in
The Handbook of Evolutionary Psychology, ed. David M.
Buss, Hoboken, NJ: Wiley, 200223.
Barsalou, Lawrence W. (2008), Grounded Cognition, Annual Review of Psychology, 59, 61745.
Boccioni, Umberto, Carlo Carra`, Luigi Russolo, Giacomo Balla,
and Gino Severini (1910), Technical Manifesto of Futurist
Painting, manifesto presented at the London exhibition, London.
Bone, Paula Fitzgerald, and Pam Scholder Ellen (1992), The Generation and Consequences of Communication-Evoked Imagery, Journal of Consumer Research, 19 (1), 93104.
Broggi, Alberto, Alexander Zelinsky, Michel Parent, and Charles
E. Thorpe (2008), Intelligent Vehicles, in Springer Handbook of Robotics, ed. Bruno Siciliano and Khatib Oussama,
New York: Springer, 117598.
Callow, Nichola, Ross Roberts, and Joanna Z. Fawkes (2006),
Effects of Dynamic and Static Imagery on Vividness of Imagery, Skiing Performance, and Confidence, Journal of Imagery Research in Sport and Physical Activity, 1 (1), 113.
Carroll, John S. (1978), The Effect of Imagining an Event on
Expectations for the Event: An Interpretation in Terms of the
Availability Heuristic, Journal of Experimental Social Psychology, 14 (1), 8896.
Chao, Linda L., and Alex Martin (2000), Representation of
Manipulable Man-Made Objects in the Dorsal Stream,
Neuroimage,12 (4), 47884.
Cian, Luca, Aradhna Krishna, and Ryan S. Elder (2014), This
Logo Moves Me: Dynamic Imagery from Static Images,
Journal of Marketing Research, 51 (2), 18497.
Clark, James, and Allan Paivio (1991), Dual Coding Theory and
Education, Educational Psychology Review, 3 (3), 149210.
Clark, Leavitt (1970), A Multidimensional Set of Rating Scales
for Television Commercials, Journal of Applied Psychology,
54 (5), 42729.
Dahl, Darren W., Rajesh V. Manchanda, and Jennifer J. Argo
(2001), Embarrassment in Consumer Purchase: The Roles
of Social Presence and Purchase Familiarity, Journal of Consumer Research, 28 (3), 47381.
DOT (US Department of Transportation, Federal Highway Administration) (2012), Manual on Uniform Traffic Control Devices, Washington, DC: Government Printing Office.
Dworkin, Bradley (2014), Chicago Swimming Pool Accident
Attorneys, http://www.personalinjuryattorney-illinois.com
Elder, Ryan S., and Aradhna Krishna (2012), The Visual Depiction Effect in Advertising: Facilitating Embodied Mental
Simulation through Product Orientation, Journal of Consumer Research, 38 (6), 9881003.
Fiske, Susan T., Daniel T. Gilbert, and Gardner Lindzey, eds.
(2010), Handbook of Social Psychology. Vol. 2, 5th ed., Hoboken, NJ: Wiley.


MT/MST by Static Images with Implied Motion, Journal of

Cognitive Neuroscience, 12 (1), 4855.
Liang, Kung-Yee, and Scott L. Zeger (1986), Longitudinal Data
Analysis Using Generalized Linear Models, Biometrika, 73
(1), 1322.
Lee, Spike W. S., and Norbert Schwarz (2012), Bidirectionality,
Mediation, and Moderation of Metaphorical Effects: The Embodiment of Social Suspicion and Fishy Smells, Journal of
Personality and Social Psychology, 103 (5), 73749
MacInnis, Deborah J., and Linda L. Price (1987) The Role of
Imagery in Information Processing: Review and Extensions,
Journal of Consumer Research, 13 (4), 47391.
Manning, C. D., S. A. Tolhurst, and P. Bawa (2012), Proprioceptive Reaction Times and Long-Latency Reflexes in Humans, Experimental Brain Research, 221 (2), 15566.
Mari, Carlo (2008), Doctoral Education and Transformative Consumer Research, Journal of Marketing Education, 30 (1),
Mazow, Leo G. (2013), Soil, Theory, and the Forum Exhibition,
American Art, 27 (3), 713.
McCloskey, Michael (1983), Naive Theories of Motion, in Mental Models, ed. Dedre Gentner and Albert L. Stevens, Hillsdale, NJ: Erlbaum, 299324.
McGill, Ann L., and Punam Anand (1989), The Effect of Vivid
Attributes on the Evaluation of Alternatives: The Role of
Differential Attention and Cognitive Elaboration, Journal of
Consumer Research, 16 (2), 18896.
Meier, Brian P., David J. Hauser, Michael D. Robinson, Chris
Kelland Friesen, and Katie Schjeldahl (2007), Whats Up
with God? Vertical Space as a Representation of the Divine,
Journal of Personality and Social Psychology, 93 (5), 699
Mick, David Glen (2008), Introduction: The Moment and Place
for a Special Issue, Journal of Consumer Research, 35 (3),
Mick, David Glen, Simone Pettigrew, Cornelia Connie Pechmann,
and Julie L. Ozanne, eds. (2011), Transformative Consumer
Research for Personal and Collective Well-Being, New York:
Morewedge, Carey K., Young Eun Huh, and Joachim Vosgerau
(2010), Thought for Food: Imagined Consumption Reduces
Actual Consumption, Science, 330 (6010), 153033.
Nairne, James S., Josefa N. S. Pandeirada, Karie J. Gregory, and
Joshua E. Van Arsdall (2009), Adaptive Memory Fitness
Relevance and the Hunter-Gatherer Mind, Psychological Science, 20 (6), 74046.
New, Joshua, Leda Cosmides, and John Tooby (2007) CategorySpecific Attention for Animals Reflects Ancestral Priorities,
Not Expertise, Proceedings of the National Academy of Sciences, 104 (42), 16598603.
Niedenthal, Paula M., Lawrence W. Barsalou, Piotr Winkielman,
Silvia Krauth-Gruber, and Francois Ric (2005), Embodiment
in Attitudes, Social Perception, and Emotion, Personality
and Social Psychology Review, 9 (3), 184211.
NiTS (Not-in-Traffic Surveillance) (2012), Traffic Safety Facts,
Washington, DC: NHTSAs National Center for Statistics and
OCraven, Kathleen M., and Nancy Kanwisher (1997), Visual
Imagery of Moving Stimuli Activates Area MT/MST, Paper
presented at the twenty-seventh annual meeting of the Society
for Neuroscience, New Orleans.
Oyserman, Daphna (2009), Identity-Based Motivation and Con-

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

Freyd, Jennifer J. (1983), Representing the Dynamics of a Static

Form, Memory and Cognition, 11 (4), 34246.
Freyd, Jennifer J.. and Ronald A. Finke (1984), Representational
Momentum, Journal of Experimental Psychology: Learning,
Memory, and Cognition, 10, (1), 12632.
Gawronski, Bertram, and B. Keith Payne, eds. (2010), Handbook
of Implicit Social Cognition: Measurement, Theory, and Applications, New York: Guilford.
Ge, Xin, Gerald Haubl, and Terry Elrod (2012), What to Say
When: Influencing Consumer Choice by Delaying the Presentation of Favorable Information, Journal of Consumer
Research, 38 (6), 100421.
Goebel, Rainer, Darius Khorram-Sefat, Lars Muckli, Hans Hacker,
and Wolf Singer (1998) The Constructive Nature of Vision:
Direct Evidence from Functional Magnetic Resonance Imaging Studies of Apparent Motion and Motion Imagery,
European Journal of Neuroscience, 10 (5), 156373.
Gonzalez, Julio, Alfonso Barros-Loscertales, Friedemann Pulvermuller, Vanessa Meseguer, Ana Sanjuan, Vicente Belloch, and
vila (2006), Reading Cinnamon Activates Olfactory
Cesar A
Brain Regions, Neuroimage, 32 (2), 90612.
Gottlieb, Jacqueline, Makoto Kusunoki, and Michael E. Goldberg
(2005), Simultaneous Representation of Saccade Targets and
Visual Onsets in Monkey Lateral Intraparietal Area, Cerebral
Cortex, 15 (8), 1198206.
Greenleaf, Eric, and Priya Raghubir (2008), Geometry in the Marketplace, in Visual Marketing: From Attention to Action, ed.
Wedel, Michel, and Rik Pieters, New York: Taylor & Francis
Group, 11342.
Gregory, W. Larry, Robert B. Cialdini, and Kathleen M. Carpenter
(1982), Self-Relevant Scenarios as Mediators of Likelihood
Estimates and Compliance: Does Imagining Make It So?
Journal of Personality and Social Psychology, 43 (1), 8999.
Hayes, Andrew F. (2013), Introduction to Mediation, Moderation,
and Conditional Process Analysis: A Regression-Based Approach, New York: Guilford.
Hubbard, Timothy L. (1995), Cognitive Representation of Motion: Evidence for Friction and Gravity Analogues, Journal
of Experimental Psychology: Learning, Memory, and Cognition, 21 (1), 24154.
Hubbard, Timothy L., and Jamshed J. Bharucha (1988) Judged
Displacement in Apparent Vertical and Horizontal Motion,
Perception and Psychophysics 44 (3), 21121.
Jiang, Yuwei, Rashmi Adaval, Yael Steinhart, and Robert S. Wyer
Jr. (2014), Imagining Yourself in the Scene: The Interactive
Effects of Goal-Driven Self-Imagery and Visual Perspectives
on Consumer Behavior, Journal of Consumer Research, 41
(2), 41835.
Kennerly, Maxwell S. (2014), Swimming Pool Accident Lawyers at the Beasley Firm, http://www.litigationandtrial.com
Kerzel, Dirk (2002), The Locus of Memory Displacement Is at
Least Partially Perceptual: Effects of Velocity, Expectation,
Friction, Memory Averaging, and Weight, Perception and
Psychophysics, 64 (4), 68092.
Koch, Christof, and Shimon Ullman (1987), Shifts in Selective
Visual Attention: Towards the Underlying Neural Circuitry,
in Matters of Intelligence, ed. Lucia M. Vaina, Dordrecht:
Reidel, 11541.
Kosslyn, Stephen M., William L. Thompson, and Giorgio Ganis
(2006), The Case for Mental Imagery, New York: Oxford
University Press.
Kourtzi, Zoe, and Nancy Kanwisher (2000), Activation in Human



Crystal Haskell, and Carrie Ballantyne (2004) The Consequences of Encoding Information on the Maintenance of Internally Generated Images and Thoughts: The Role of Meaning Complexes, Consciousness and Cognition, 13 (4),
Smith, Eliot R., and Gun R. Semin (2004), Socially Situated
Cognition: Cognition in Its Social Context, Advances in Experimental Social Psychology, 36, 53117.
(2007), Situated Social Cognition, Current Directions
in Psychological Science, 16 (3), 13235.
Spencer, Steven J., Mark P. Zanna, and Geoffrey T. Fong (2005),
Establishing a Causal Chain: Why Experiments Are Often
More Effective than Mediational Analyses in Examining Psychological Processes, Journal of Personality and Social Psychology, 89 (6), 84551.
Taylor, Shelley E., and Sherry K. Schneider (1989) Coping and
the Simulation of Events, Social Cognition, 7 (2), 17494.
Tschida, Stephen (2013), AAA Study: PedestrianParking Lot
Accidents on the Rise, http://www.wjla.com/articles/2013/
Wang, Kai-Yu, and Laura A. Peracchio (2007), Reading Picture:
Understanding the Stylistic Properties of Advertising Images, in Go Figure! New Directions in Advertising Rhetoric,
ed. Edward F. McQuarrie and Barbara J. Phillips, Armonk,
NY: M. E. Sharpe, 20526.
Wenzlaff, Richard M., and Daniel M. Wegner (2000), Thought
Suppression, Annual Review of Psychology, 51 (1), 5991.
Wood, Stacy, Melayne Morgan McInnes, and David A. Norton
(2011), The Bad Thing about Good Games: The Relationship
between Close Sporting Events and Game-Day Traffic Fatalities, Journal of Consumer Research, 38, (4), 61121.

Downloaded from http://jcr.oxfordjournals.org/ by guest on August 13, 2016

sumer Behavior, Journal of Consumer Psychology, 19 (3),

Paivio, Allan (1969), Mental Imagery in Associative Learning
and Memory, Psychological Review, 76 (3), 24163.
Pashler, Harold (1998), The Psychology of Attention, Cambridge,
MA: MIT Press.
Patrick, Vanessa M., and Laura A. Peracchio (2010) Curating
the JCP Special Issue on Aesthetics in Consumer Psychology:
An Introduction to the Aesthetics Issue, Journal of Consumer
Psychology, 20 (4), 39397.
Petrova, Petia K., and Robert B. Cialdini (2005), Fluency of Consumption Imagery and the Backfire Effects of Imagery Appeals, Journal of Consumer Research, 32 (December),
(2008), Evoking the Imagination as a Strategy of Influence, in Handbook of Consumer Psychology, ed. Curt Haugtvedt, Paul Herr, and Frank Kardes, Mahwah, NJ: Erlbaum,
Pratt, Jay, Petre V. Radulescu, Ruo Mu Guo, and Richard A.
Abrams (2010), Its Alive! Animate Motion Captures Visual
Attention, Psychological Science, 21 (11), 172430.
Ptak, Radek, and Julia Fellrath (2013), Spatial Neglect and the
Neural Coding of Attentional Priority, Neuroscience and
Biobehavioral Reviews, 37 (4), 70522.
Ratcliff, Roger (1993), Methods for Dealing with Reaction Time
Outliers, Psychological Bulletin, 114 (3), 51032.
Robinson, Michael D. (2007), Lives Lived in Milliseconds: Using
Cognitive Methods in Personality Research, in Handbook of
Research Methods in Personality Psychology, ed. Richard W.
Robins, R. Chris Fraley, and Robert F. Krueger, New York:
Guilford, 34559.
Smallwood, Jonathan, Rory C. OConnor, Megan V. Sudberry,